[Fix] gather_gemv small fix (#452)

Author: Sibylau

tritonbench/operators/gather_gemv/operator.py

@@ -38,7 +38,7 @@ def __init__(
     ):
         super().__init__(tb_args, extra_args)
 
-    @register_benchmark(baseline=True)
+    @register_benchmark()
     def test_0(self, p1, p2, p3) -> Callable:
         return lambda: triton_test_0(p1, p2, p3)
 

tritonbench/operators/gather_gemv/triton_gather_gemv.py

@@ -126,4 +126,4 @@ def triton_gemv_0(arg0_1, arg1_1, arg2_1):
 
         grid = lambda META: (triton.cdiv(2 * S, META["XBLOCK"]),)
         triton_red_fused_mv_0[grid](arg1_1, arg0_1, arg2_1, buf1, xnumel, rnumel)
-    return (reinterpret_tensor(buf1, (2, S), (S, 1), 0),)
+    return reinterpret_tensor(buf1, (2, S), (S, 1), 0)